Nature Materials

Directed cell migration underlies many biological phenomena, from embryonic development to tumor metastasis and organ fibrosis. Most cells typically migrate toward stiffer regions of their extracellular matrix -a behavior known as positive durotaxis. Here we show that culture on rigid plastic reinforces this response, whereas preconditioning in soft 3D physiomimetic environments reprograms migration towards softer environments, a phenomenon known as negative durotaxis. Fetal rat lung fibroblasts preconditioned in 3D physiomimetic hydrogels exhibited negative durotaxis and accumulated near [~]5 kPa, corresponding to the physiological stiffness of the lung. In contrast, genetically identical cells maintained on conventional 2D plastic substrates migrated up stiffness gradients, toward stiffer regions. Although both populations displayed a biphasic force-stiffness relationship, they differed in force magnitude and cytoskeletal organization. Molecular-clutch modeling revealed that durotaxis reversal emerges from two distinct mechanical regimes: a mechanosensitive, high-motor-clutch state that stabilizes adhesions on stiff substrates and drives positive durotaxis, and a low-motor, weak-adhesion state in which clutch slippage on the stiff side causes negative durotaxis. Our results show that durotaxis direction is not an intrinsic cellular property. Rather, it emerges from the interplay between motor activity and adhesion dynamics and can be tuned by culture conditions.

Engineered living materials (ELMs) promise genetically programmable functions by coupling biological regulation to synthetic material responses. Here, we introduce genetically encoded, reversible shape-morphing in a peptide-crosslinked polyethylene glycol (PEG) hydrogel whose network density is modulated by opposing enzymatic pairs that induce crosslinking or hydrolysis. This molecular programmability alternates the hydrogel between deswelling and swelling/disintegration and produces 2 - 5-fold changes in mechanical properties. By fabricating a bilayer hydrogel with an inert layer, these molecular modulations are translated into a reversible and directional motion with angular bending motions exceeding 80{degrees}. Further, by embedding genetically engineered bacteria or interfacing mammalian cells, producing the relevant enzymatic cues, the reversible shape-morphing of these ELMs is programmed at the genetic level. We further demonstrate genetically programmed, autonomous reversible bending in a bilayer hydrogel controlled by out-of-equilibrium counteracting biochemical reactions with dynamically changing respective reaction rates. This work establishes a concept where coordinated polymer/peptide material engineering and synthetic biology yield autonomous shape-morphing ELMs, opening avenues toward biohybrid soft robotics, adaptive microfluidic systems, and dynamic biomedical interfaces.

Collective cell migration drives tissue morphogenesis, repair and remodeling, and is often accompanied by transitions from solid-like to fluid-like states. While such tissue fluidization has been linked to physical parameters such as cell density, shape and activity, how it is actively regulated by mechano-chemical interplay remains unclear. Previous research has shown that transient attenuation of actomyosin contractility induces a transition from pulsatile, spatially confined motion to coherent, persistent long-range collective flow; however, the underlying cellular and signaling mechanisms remain unclear. Here we uncover the mechanistic basis by which transient perturbation of cell contractility reprograms the migration mode of confluent epithelial cells into a leader-like, fluidizing state, by combining kinase-reporter live imaging, force measurements and mathematical modeling. This transition arises from coordinated changes in cell morphology, mechanics, and signaling, including reduced cortical tension, enhanced cell-substrate adhesion and traction forces, and increased tissue deformability. At the signaling level, this process is accompanied by a rewiring of extracellular signal-regulated kinase (ERK)-mediated mechanotransduction toward a protrusion-coupled mode that sustains migration even under fully confluent conditions. Consistently, a multicellular computational model further demonstrates that protrusion-driven migration is sufficient to promote shape-velocity alignment and drive a transition from caged to flocking-like collective states. Together, our results identify transient mechanical relaxation as a trigger for an intrinsic leader-like state that fluidizes epithelial confluent tissues through coordinated remodeling of cytoskeletal, adhesive, and signaling systems.

Altered mechanical properties of the tumor microenvironment (TME) influence cancer progression, yet the mechanistic basis by which 3D mechanics shape CAF heterogeneity and downstream tumor drug response remains poorly understood. Here, we engineered a modulus-tunable gelatin methacryloyl (GelMA) hydrogel platform spanning a normal-like (soft [~]2 kPa) to desmoplastic-like (stiff[~]40 kPa) range to culture primary breast CAFs under 3D confinement. CAFs exhibited pronounced volumetric morphoadaptation across matrices, with soft 3D matrices supporting larger, more protrusive morphologies and stiff gels constraining cell geometry. In contrast to canonical 2D paradigms, nuclear YAP localization was reduced in stiff 3D matrices and varied substantially across cells, consistent with a dominant role for 3D geometric/volumetric state in regulating mechanotransduction. Functionally, in transwell co-culture with MCF-7 spheroids under paclitaxel treatment, CAFs cultured in stiff 3D matrices induced a broader chemoresistance-associated transcriptional program, whereas soft 3D matrices CAFs favored stress/checkpoint-like responses. A 2D monolayer comparator indicated that coordinated resistance-associated programs emerge most clearly in 3D tumor architecture. Together, these results establish a GelMA-based biomaterials framework in which CAF volumetric state provides a quantifiable intermediate linking 3D matrix mechanics to mechanotransduction and tumor drug-response programs, motivating future strategies to modulate stromal function through mechanically controlled cell-state regulation.

The mechanical toughness of bone and teeth relies on residual stresses generated during mineralisation, where the dehydration of collagen fibrils leads to contraction, putting the mineral phase under compression. While macroscopic stiffening of collagen upon drying is well-documented, the atomic-level structural rearrangements driving this phenomenon have remained elusive. By performing molecular dynamics simulations, we demonstrate that collagen contraction is not homogeneous but is driven by specific charged motifs. We identify a critical sequence-dependent rule for contraction: oppositely charged side chains must be separated by at least four residues to drive backbone contraction. While salt bridges can form between side chains at a distance less than four residues without perturbing the helix, those at greater distances cannot form without rupturing backbone hydrogen bonds. Consequently, dehydration forces these distant charges together, breaking local backbone structure and driving collagen contraction. These findings imply that collagen sequences are evolutionarily tuned to actively control tissue mechanics and redefines collagen as an active mechanical element rather than a passive scaffold. Furthermore, this framework provides a molecular basis for understanding mechanical failure associated with pathologies and ageing, while simultaneously opening avenues for designing bio-inspired materials with tunable pre-stress and fracture resistance.

The spatial arrangement of cells is fundamental to the mechanical and functional integrity of three-dimensional (3D) tissues, yet engineering spatially well-controlled tissue architectures remains challenging. Here, we computationally investigated how layered tissue architectures can be designed by modulating cell-cell interfacial tension. We performed simulations using a 3D vertex model and systematically varied interfacial tension magnitudes. The simulations generated a range of layered tissue architectures, including planar monolayers, bilayers, and structurally stratified states. In homogeneous cell populations, increasing interfacial tension drove transitions from monolayer to structurally stratified configurations. In heterogeneous populations, differential interfacial tensions induced out-of-plane cell sorting and the formation of compositionally sorted multilayers. Moreover, a recursive tension design strategy enabled hierarchical organization of multiple cell types into separate layers. Notably, this recursive scheme uses only two tension levels (high vs. low) assigned across interfaces and can, in principle, be extended to specify layered architectures with an arbitrary number of layers. Together, these results identify cell-cell interfacial tension as a tunable mechanical parameter for regulating layered tissue architecture and provide design principles for layered tissue engineering and regenerative medicine.

Neuroelectronic interfaces hold great promise to restore functions in neurological disorders or motor dysfunctions, but current devices struggle to integrate seamlessly within living tissues. Here we report a transformative approach to create bionic neurons that autonomously build integrated fluorescent fibrils and demonstrate their role as neuromodulators. Using a combination of cell biology, ultrastructural, imaging and nanospectroscopical approaches, we deciphered the unique biosynthetic pathway employed by the cells to self-fabricate these nanoelectronics and uncover their hybrid structure. Importantly, patch clamp recordings revealed their neuromodulatory potential, through the perturbation of membrane electrical properties and the early rising phase of the action potential. Deciphering how basic molecular elements self-organize into complex architectures within biological environments could unlock the ability to engineer natural electroactive systems directly inside living organisms. This capability could be used to create conductive pathways between arbitrarily defined neurons, microcircuits, or nervous system regions, effectively writing connections into living brains.

Compartmentalization is a defining feature of cellular systems, yet how early compartments could undergo repeated cycles of growth, division, and content organization without complex chemistry remains unresolved. Here we study a minimal membrane-based system subjected to periodic hydration- dehydration cycles, mimicking fluctuating physical environments on the early Earth. We show that cyclic environmental conditions alone drive a sequence of reproducible compartment dynamics, including macromolecule encapsulation, membrane growth, division, and the generation of a highly crowded interior. These processes emerge from biophysical transformations of a single-component membrane and do not require any chemical reactions or metabolic activity. Importantly, compartments retain their structural integrity across multiple cycles, enabling repeated encapsulation without loss of individuality. Our results demonstrate that fluctuating physical conditions can be transduced by membrane biophysics into sustained, cell-like cycles, challenging the view that primordial cellular dynamics necessarily required chemically driven growth and division.

Fibrotic responses at biomaterial-tissue interfaces limit implant integration and regenerative healing, yet how the interaction between biomaterials and the extracellular matrix (ECM) regulates fibroblast activation remains poorly understood. Granular hydrogels including microporous annealed particle scaffolds (MAP) reduce fibrosis, while chemically and mechanically matched hydrogels do not, suggesting a dominant role for scaffold architecture. In this model, MAP scaffolds allow collagen infiltration and form physically continuous composites, whereas hydrogels exclude collagen and generate interfacial slip planes. To isolate how biomaterial architecture influences extracellular matrix (ECM) integration and fibroblast activation, we developed a reductionist in vitro model that integrates collagen type I with either microporous annealed particle (MAP) scaffolds or chemically and mechanically matched bulk hydrogels. This physical integration stabilizes collagen architecture, limits fibroblast-mediated matrix compaction, suppresses contractility, and attenuates myofibroblast transition. Fibroblasts in mechanically integrated environments exhibit reduced expression and nuclear localization of NF-{kappa}B and are enriched for quiescent phenotypes. Together, these findings identify biomaterial-ECM physical continuity as a design principle for limiting fibrotic signaling.

Cancer cells breach the extracellular matrix (ECM) using both protease-driven degradation and force-driven physical remodeling, yet most anti-metastatic drug screens still rely on biochemical assays that overlook cell-matrix mechanical reciprocity. Here, we present a fully synthetic 3D invasion platform based on cellular force-responsive polyisocyanide (PIC) hydrogels that isolates biophysical invasion mechanisms. Cell-generated forces align and densify the PIC fibrous network, reproducing hallmark matrix remodeling seen in the tumor microenvironment. A constitutive model, parameterized by the critical stress for strain stiffening effect, links matrix nonlinear elasticity to pericellular stiffening, long-range mechanotransmission, and intercellular coupling. Using this system, we show that breast cancer cells invade by pulling and pushing the network even when matrix metalloproteinases are inhibited, revealing a physical bypass of protease blockade. Accordingly, broad-spectrum metalloproteinase inhibitors that suppress invasion in Matrigel fail to inhibit invasion here, exposing a limitation of current drug-evaluation pipelines. In co-culture, cancer-associated fibroblasts markedly accelerate invasion by generating aligned fiber tracks through higher contractility, implicating CAF-driven mechanical remodeling as a key route for breaching barriers during metastasis. The platform is thermoresponsive, compatible with standard Transwell formats, enables direct imaging of fiber architecture and invasion fronts, and decouples biophysical from biochemical cues for mechanism-aware, animal-free assessment of anti-metastatic therapies.

Synthetic living constructs, which lack the long histories of selection in ecological contexts that shape behaviors of conventional organisms, offer an important complement to traditional studies of learning. Could novel biobots exhibit sensing and memory of experiences? Here, we investigated the effects of chemical stimuli on basal Xenobots - autonomously motile entities derived from Xenopus embryonic ectodermal explants (with no additional sculpting or bioengineering). We quantified and characterized the coordinated ciliary activity that generates fluid flow fields guiding the trajectory of Xenobot motion. We also show distinct and specific changes in Xenobot behavior after brief exposure to Xenopus embryonic cell extract and to ATP. These two experiences produced distinct, long-term, stimulus-specific memories, detectable through both transcriptional and physiological signatures. Exposure to specific environmental stimuli induced alterations in the spatiotemporal patterns of calcium signaling across Xenobots. Together, these data lay a foundation for characterizing the capabilities of synthetic cellular collectives to sense and discriminate among stimuli, as well as store functional information in a non-neural context. Understanding behavioral competencies in novel, non-neural systems have broad implications across evolutionary biology, behavioral science, bioengineering, and bio/hybrid robotics.

The yield stress at which biomaterials undergo plastic deformation limits the stresses that can be developed in encapsulated growing tissues. While mechanical properties of the matrix such as stiffness and viscoelasticity have a profound effect on cells, the role of yield stress has remained challenging to define. Here we design a self-healing granular hydrogel platform with supramolecular host-guest dynamic crosslinkers to precisely and quantitatively tune the stress at which the matrix repeatedly yields and reconfigures around tissues as they grow. Designed to provide similar mechanical constraints as a mesh stress ball, matrix yield stresses can be tuned between 12 and 370 Pa, while maintaining a storage modulus below [~]0.1kPa. We show that this range of yield stress is sufficient to promote or limit peripheral shedding in a model of non-adhesive cancer migration; and that early development of midbrain organoids is exquisitely sensitive to yield stress. Optimal yield stresses of only 25 Pa promoted budlike protrusions and large, lumenized neural rosettes, while variations as small as 10 Pa limited these phenotypes. These studies demonstrate that morphogenesis and tissue organization are exquisitely sensitive to yield stress, suggesting a new material property to target in designing biomaterials for disease modeling and regenerative medicine.

Three-dimensional (3D) stem cell-based cultures have emerged as promising in vitro model systems for studying human neurodevelopment. Current neural organoid protocols lack well-defined extracellular matrix (ECM) signaling and are limited by the formation of irregular tissue morphologies with multiple organizing centers, in contrast to the single neuroepithelial structure that emerges during embryonic development. This variability limits inter-organoid reproducibility and constrains their utility for modeling early developmental processes. To overcome these limitations, we leverage a materials-based approach to impose dynamic boundary conditions that extrinsically guide the self-organization of human induced pluripotent stem cells (iPSCs). Specifically, we develop a family of hyaluronic acid-elastin-like protein (HELP) hydrogels crosslinked with dynamic covalent bonds that recapitulate key biochemical and biophysical properties of the developing human neural ECM. Within these HELP hydrogels, iPSCs robustly self-organize from a single cell into complex neuroepithelial tissues with a single lumen. By tuning the boundary conditions imposed by the hydrogel, we identify matrix stress relaxation rate and tensional homeostasis as key regulators of single-lumen rosette formation and maintenance. With this hydrogel-enabled system, we identify phenotypic abnormalities in an early neurodevelopmental model of 22q11.2 deletion syndrome. Ultimately, our tunable engineered hydrogel supports the initiation of single-cell derived 3D neuroepithelial tissues, enables investigation into how matrix-imposed boundary conditions guide developmental morphogenesis, and establishes a reproducible platform for disease modeling.

Cell alignment is a fundamental process in tissue morphogenesis. While density-dependent collective cell alignment has been widely observed, its underlying mechanisms remain poorly understood. Here, using C2C12 myoblasts, we demonstrate that static uniaxial mechanical stretch induces collective cell alignment in a density-dependent manner: densely populated cultures align robustly, whereas sparse populations do not. We reveal a biphasic alignment process, comprising an initial passive phase and a subsequent active phase. The passive phase, driven by substrate deformation, transiently biases cell orientation along the stretch axis regardless of density. In the active phase, initial alignment progressively dissipates in low-density cultures, but is sustained and reinforced in high-density cultures. Supported by coarse-grained agent-based simulations, we propose that self-generated cellular forces facilitate kinetic transitions between orientations, enabling cells to explore orientational states, whereas cell-cell interactions provide a thermodynamic bias that stabilizes the locally aligned state. In dense cultures, strong intercellular interactions promote this stabilization, enabling persistent alignment. In contrast, sparse cultures lack sufficient cell-cell interaction, leading to alignment dissipation. Within this C2C12 system, our findings highlight the cooperative roles of cellular forces and intercellular interactions in orchestrating multicellular ordering, offering new insights into mechanobiology of tissue morphogenesis.

Cells can experience time-varying mechanical cues, particularly when navigating through changing and complex microenvironments. Yet whether and how cells retain and use a short-term mechanical memory of recent deformations remains unclear. Here we show that, in glioblastoma cells, this memory is encoded by transient cytoskeletal anisotropy. Using uniaxial magneto-mechanical actuation aligned or perpendicular to the cell long axis, nanoindentation, and selective cytoskeletal perturbations, we find that distinct architectures of the actin cytoskeleton drive opposite mechanical responses: actin stress fibers mediate stiffening under stretch, whereas the actin cortex underlies softening under perpendicular loading. Vimentin intermediate filaments are essential to stabilize actin organization under load, preserving deformation-specific mechanics. Quantitative imaging reveals that mechanical actuation induces network-specific alignment and anisotropy, stronger for actin than vimentin, that persists transiently after unloading and bias subsequent responses, revealing a short-lived, deformation-dependent mechanical memory. To integrate these observations, we develop a multi-network constitutive model that links cytoskeletal architecture and loading history to cell-scale mechanics, reproducing both the asymmetric mechanical responses and the measured reorganization dynamics. These findings provide a structural basis for short-term mechanical memory and suggest how cancer cells could exploit residual anisotropy to adapt to fluctuating solid stresses and confinement, transiently biasing polarization, force transmission, and directional persistence during invasion. They also identify vimentin-actin coupling and the kinetics of cytoskeletal remodeling as potential levers to limit the mechanical adaptability of invasive cancer cells.

Understanding protein phase separation in cellular environments remains a major challenge, as ex vivo assays often fail to capture the influence of environmental context - such as crowding, multimodal interactions, and the dynamic properties of the cytosol or nucleus. Here, we introduce programmable DNA-based protonuclei (PN) as nucleus-mimicking compartments to probe phase separation of the neurodegeneration-linked protein FUS. We show that FUS partitioning and condensate formation are highly sensitive to nucleic acid sequence, spatial confinement, and viscoelastic properties of the PN core. Notably, classical test-tube affinity assays fail to predict protein behavior within the crowded and multivalent PN environment. By tuning DNA crosslinking, we modulate condensate dynamics and suppress liquid-to-solid transitions of FUS - a hallmark of disease. These findings demonstrate that multivalent, confined environments fundamentally reshape protein-nucleic acid interactions and phase behavior. The PN platform complements test-tube assays and complex cellular settings and enables to dissect nuclear condensates under controllable conditions.

A central problem in soft and biological physics is how molecular-scale activity and remodelling coarse-grain into emergent mechanical laws at larger scales. In growing cell walls (polymeric composite materials that surround 90% of living organisms cells) irreversible deformation is not controlled by elastic stress alone. Instead, growth depends on the interplay between energy storage, dissipation, and the local timing of viscoelastic relaxation. Although dynamic atomic force microscopy (AFM) resolves storage and loss moduli (E', E'') of living walls at nanometre resolution, these observables have remained phenomenological and disconnected from constitutive field variables. Here we introduce a physics-based inversion framework that converts AFM measurements of epidermal cells of living Arabidopsis plants into spatially resolved fields of stiffness k, viscosity , and relaxation time{tau} . By analysing the spatial gradients of E' and E'', we uncover organized mechanical heterogeneities governed by cellular confinement and stress focusing. We demonstrate that the local relaxation time is encoded directly in the coupling between storage and dissipation, yielding the pointwise relation{tau} = (1/{omega}) {partial}E/{partial}E, where{omega} is the indentation frequency. This relation enables model-independent extraction of mechanical timescales and establishes a general route from nanoscale non-equilibrium rheology to continuum descriptions of growth in living and active soft materials. SignificanceHow molecular-scale activity gives rise to tissue-scale form is a central challenge in biological physics. Although growth is fundamentally a non-equilibrium mechanical process, experimental measurements at the nanoscale have not been directly connected to the constitutive parameters that govern morphogenesis. We introduce a framework that converts dynamic atomic force microscopy maps of storage and loss moduli into spatially resolved fields of stiffness, viscosity, and relaxation time in living cell walls. By revealing that mechanical relaxation is encoded in the local coupling between elastic storage and viscous dissipation, our work provides a route from nanoscale rheology to growth-relevant mechanical timing. This establishes a quantitative bridge between molecular remodeling and continuum mechanics, enabling direct experimental constraints on multiscale theories of morphogenesis.

Curli fibers produced by Escherichia coli are functional amyloids that activate Toll-like receptor 2 (TLR2), initiating innate immune responses at mucosal surfaces. While microbiome-derived curli contribute to host-microbe interactions, their intrinsic immunostimulatory activity limits their utility as programmable scaffolds for engineered probiotic systems, and dysregulated TLR2 activation has been linked to inflammatory bowel disease, systemic lupus erythematosus, neurodegeneration, and sepsis. Here, we engineered E. coli Nissle 1917 to produce modified curli fibers designed to inhibit TLR2 through two mechanistically distinct strategies: steric shielding via silk-elastin-like protein sequences, and direct receptor antagonism via a known TLR2 antagonist, staphylococcal superantigen-like protein 3 (SSL3). Both engineered variants assembled into structurally intact amyloid fibers and exhibited significantly reduced intrinsic TLR2-dependent NF-{kappa}B activation in reporter cells. In competitive inhibition assays against structurally diverse TLR2 agonists, the SSL3 fusion achieved near-complete inhibition maintained under rising agonist load, while steric shielding provided moderate, agonist-class-dependent inhibition. In primary human monocyte-derived dendritic cells, the SSL3 fusion robustly attenuated IL-8 secretion and transcriptional induction of IL-8, IL-6, and IL-1{beta}, whereas steric shielding produced only partial attenuation that did not translate to broad inflammatory suppression. These results establish engineered curli as a tunable platform for receptor-specific modulation of innate immune signaling and highlight the broader potential of modular microbial amyloids as programmable interfaces for engineering host-microbe interactions at mucosal surfaces. IMPORTANCEBacteria residing in the gut produce protein fibers called curli that potently activate the immune system through a receptor called Toll-like receptor 2 (TLR2). While TLR2 plays a beneficial role in maintaining gut health, its overactivation drives chronic inflammation in conditions including inflammatory bowel disease, autoimmune diseases, neurodegenerative diseases, and sepsis, and curli fibers have been directly implicated in several of these conditions. Here, we engineered curli fibers produced by the probiotic E. coli Nissle 1917 to inhibit TLR2 activation, transforming a naturally inflammatory bacterial fiber into a programmable immune modulator. We demonstrated that direct receptor antagonism, rather than steric shielding, is required for effective immune modulation in primary human immune cells, establishing a design principle for engineering bacteria-derived fibers as programmable interfaces with host immunity. The modularity of the curli scaffold positions this platform as a broader tool for programming interactions between probiotic bacteria and the mucosal immune system.

Microbial functional amyloids play central roles in biofilm formation and serve as foundational building blocks for autogenic engineered living materials (ELM), yet the structural design space governing their assembly and stability remains poorly defined. In Escherichia coli, the {beta}-solenoid protein CsgA functions as a canonical extracellular matrix scaffold, but prior engineering efforts have primarily focused on terminal functionalization rather than modification of the {beta}-solenoid core itself. Here, inspired by the evolutionary diversification of CsgA-like proteins, which expands along the vertical fiber axis, we explore a second orthogonal axis of structural plasticity: the horizontal dimension of the {beta}-solenoid. We rationally designed a library of CsgA variants in which the length of individual {beta}-strands was systematically reduced or extended from the native seven residues to lengths spanning 3-21 residues, while preserving conserved gate residues and loop regions. Integration of AI-based structure prediction using AlphaFold2 with all-atom molecular dynamics simulations reveals that {beta}-solenoid stability arises from a balance among strand length, residue composition, and solvent interactions, thereby defining both lower and optimal bounds for nanofiber assembly. Experimental validation demonstrates that engineered Escherichia coli can secrete and assemble these CsgA variants into extracellular nanofibers through the native curli biogenesis machinery, while preserving the characteristic cross-{beta} architecture. Additionally, the CsgA {beta}-solenoid variant library translates molecular design into macroscopic ELM, with deletion variants showing an inverse relationship between stiffness and extensibility, from highly extensible 3aa to stiff, strong 5aa films. Insertion-based variants largely retain CsgA-like extensibility while enabling tunable stiffness and strength across strand lengths. Together, these findings uncover previously unrecognized structural plasticity in microbial {beta}-solenoid proteins and establish {beta}-strand length as a generalizable design parameter linking molecular architecture to nanofiber stability, with implications spanning microbial functional amyloids and the rational design of autogenic engineered living materials.

The traditional view of protein self-assembly posits a binary choice between phase separation into fluid condensates and nucleation into crystalline amyloid fibers. However, this framework is incomplete. Experiments show that liquid condensates are non-equilibrium systems that mature into solid-like structures mediated by amphiphilic prion-like domains (PLDs). Being spatially organized yet dynamic, lyotropic phases represent an intermediate regime between these states. Using physics-based de novo protein design (Evo-MD), we identify a vast amphiphilic motif space encoding fluid lyotropic phases (e.g., micelles and bicelles). TEM, CD, and AlphaFold predictions confirm that these motifs also assemble into amyloid-based hydrogels as thermodynamic endpoints. Notably, the molecular grammar of lyotropic motifs overlaps strongly with that of PLDs and LARKS. Thus, while PLDs likely evolved to stabilize condensates through transient interactions near criticality, our results show that these same amphiphilic forces inherently encode lyotropic structuring and subsequent amyloid formation - linking functional condensation with pathological aggregation.